A genome-wide approach reveals novel imprinted genes expressed in the human placenta

Abstract

Genomic imprinting characterizes genes with a monoallelic expression, which is dependent on the parental origin of each allele. Approximately 150 imprinted genes are known to date, in humans and mice but, though computational searches have tried to extract intrinsic characteristics of these genes to identify new ones, the existing list is probably far from being comprehensive. We used a high-throughput strategy by diverting the classical use of genotyping microarrays to compare the genotypes of mRNA/cDNA vs. genomic DNA to identify new genes presenting monoallelic expression, starting from human placental material. After filtering of data, we obtained a list of 1,082 putative candidate monoallelic SNPs located in more than one hundred candidate genes. Among these, we found known imprinted genes, such as IPW, GRB10, INPP5F and ZNF597, which contribute to validate the approach. We also explored some likely candidates of our list and identified seven new imprinted genes, including ZFAT, ZFAT-AS1, GLIS3, NTM, MAGI2, ZC3H12Cand LIN28B, four of which encode zinc finger transcription factors. They are, however, not imprinted in the mouse placenta, except for Magi2. We analyzed in more details the ZFAT gene, which is paternally expressed in the placenta (as ZFAT-AS1, a non-coding antisense RNA) but biallelic in other tissues. The ZFAT protein is expressed in endothelial cells, as well as in syncytiotrophoblasts. The expression of this gene is, moreover, downregulated in placentas from complicated pregnancies. With this work we increase by about 10% the number of known imprinted genes in humans.

Figure 1. Sequencing results of SNPs in the human ZFAT and ZFAT-AS1 genes. Imprinted expression was deduced by comparing genotypes of gDNA, cDNA and maternal gDNA, in placenta using SNPs rs rs3739423 (A), rs894344 (B) and rs894343 (C) and in lymphocytes (D).

Figure 2. Sequencing results of SNPs in the murine (A) and bovine (B) Zfat genes. Imprinted expression was deduced by comparing genotypes of gDNA, cDNA and maternal gDNA, in placenta and other tissues.

Figure 3. Sequencing results of SNPs in the murine Magi2 gene. Imprinted expression was deduced by comparing genotypes of gDNA, cDNA and maternal gDNA, in placenta and other tissues.

Figure 4. Real-time RT-PCR of ZFAT in human placentas. Expression of the ZFAT gene was compared between Controls (n = 17), IUGR (pregnancies with intrauterine growth restriction) (n = 16), PE (pregnancies complicated with preeclampsia) (n = 15) and PE + IUGR (n = 5), after normalization by the SDHA housekeeping gene. *p = 0,02, ** p = 0,002 in a group vs. the control group.

Figure 5. Immunohistochemistry to detect ZFAT on sections of human placentas. (A and B) (19 and 32 weeks of amenorrhea respectively) and (C) (fetal lung). (+) shows the detection with the primary antibody sc-87510 and (-) the negative control of detection without this antibody. Plain arrows show the strong labeling of endothelial cells while dotted arrows point to the syncytiotrophoblast labeling. (D) A western blot of human and mouse tissues revealed by an anti-ZFAT antibody. PBL stands for peripheral blood leukocytes whereas JEG-3 is a human choriocarcinoma cell line commonly used as a model of trophoblast.

Figure 6. Cloning and sequencing of PCR fragments obtained from bisulfite-treated placental gDNA in the promoter CpG island of the LIN28B gene. Black and white circles represent methylated and unmethylated cytosines respectively over the 32 CpG dinucleotides under study. SNPs are present within this fragment and allow the study of the allelic segregation and parental transmission.